non-destructive filter not OK?

[Dsbbw2020]

Hello,

Description of the problem:

I want to use the recipe for "Freaky Details" (https://www.gimp-forum.net/Thread-Look-m...-Gimp-2-10), That is:

 - Make two copies of your original image (three layers total)
 - Color>Invert the top layer
 - Filters>Blur>Selective Gaussian blur the top layer (radius around 5 px, max delta around 0.2)
 - Add a layer group at the top, and move the top two layers in it
 - Set the color-inverted layer to Vivid light mode
 - Set the layer group to Overlay mode

If I do so in Gimp3.0 RC1 everything is OK

Next I change the original image with the function: color -> exposure (+ or - , both tested)
After applying the recipe the result is now unexpected. Sie screen1.jpg.

If I remove the non-destructive filter (exposure) from the top image (the inverted one) everything is OK.
See screen2.jpg.

   

   

[Ofnuts]

(11-14-2024, 09:58 AM)Dsbbw2020 Wrote: Hello,

Description of the problem:

I want to use the recipe for "Freaky Details" (https://www.gimp-forum.net/Thread-Look-m...-Gimp-2-10), That is:

 - Make two copies of your original image (three layers total)
 - Color>Invert the top layer
 - Filters>Blur>Selective Gaussian blur the top layer (radius around 5 px, max delta around 0.2)
 - Add a layer group at the top, and move the top two layers in it
 - Set the color-inverted layer to Vivid light mode
 - Set the layer group to Overlay mode

If I do so in Gimp3.0 RC1 everything is OK

Next I change the original image with the function: color -> exposure (+ or - , both tested)
After applying the recipe the result is now unexpected. Sie screen1.jpg.

If I remove the non-destructive filter (exposure) from the top image (the inverted one) everything is OK.
See screen2.jpg.

Works for me.... Do you set the exposure to the same value in all three copies?

An interesting question: Color Invert is always applied, so is it applied after the exposure effect or before? And if before, shouldn't the exposure be applied in the opposite direction? (amount being left as an exercise for the reader).

To behave correctly this should be a single GEGL graph, with the same input applied to 3 different places. I wondered about this a while back, and didn't find a solution at the time.

[rich2005]

I think I see the issue.

   

In (1) The GEGL filter (exposure) is applied, Followed by the procedure - duplicate etc. Each duplicate has that filter attached. And that gives one result

In (2) the same GEGL exposure applied then merged with the image. The procedure applied and that is not the same effect as (1).

Is there a fix, apart from (2) I do not know.

EDIT: Just a thought. Apply the filter to the layer group as: (different computer, still Gimp 3 )  That is adjustable.